Virtual reality (VR) cameras have multiple component cameras (for example, eight component cameras) that cover the entire three-dimensional (3D) field of view around themselves, and every component camera has its own image pipeline which processes raw images from the respective component camera to obtain a quality image. The images obtained from multiple component cameras are then ‘stitched’ for virtual reality consumption. The stitching is similar to panorama algorithms and is made possible, as a VR camera rig is so designed that the component cameras have overlapping field of views (FOVs).
In a scenario, where a VR camera is moving, for example worn by a hiker on a backpack or on a dolly for shooting a VR movie, the component cameras keep seeing different scenes, for example, the FOVs of the component cameras keep changing. Existing algorithms for adjusting 3A parameters (e.g., exposure, focus and white balance) of the component cameras take time to adapt and converge, similar to any conventional cameras. When one suddenly changes the FOV of a conventional camera, the color and exposure is initially ‘off’, and in a matter of a few seconds or within a second, it corrects itself. While the above is tolerable in conventional video, in the case of VR, when one is immersed in the content, if the image content varies as one moves his/her head, it may be very discomforting.
In an example, imagine a room with a window and the VR camera, in which a camera A may be imaging the scene near the window and outside, while an adjacent camera, say, camera B is imaging an interior part of the room. The exposures for cameras A and B will be very different, with the camera A having a very low exposure and the camera B having a moderately high exposure. If the cameras A and B move such that the part of the scene containing the window comes into the FOV of camera B, then, the video captured by the camera B will be saturated and overly bright. This is because it takes some time for the camera 3A algorithms to figure out that the scene has changed, and exposure has to be accordingly adjusted. When content like this is consumed via a VR headset in an immersive manner, this may cause discomfort and lead to a sub-par user experience.